VOL. 53, NO. 3
JOURNAL OF THE ATMOSPHERIC SCIENCES
1 FEBRUARY 1996
Multispectral, High-Resolution Satellite Observations
of Plumes on Top of Convective Storms
VINCENZO LEVIZZANI
Institute FISBAT–CNR, Bologna, Italy
MARTIN SETVÁK
Satellite and Radar Department, Czech Hydrometeorological Institute, Prague, Czech Republic
(Manuscript received 1 November 1994, in final form 18 August 1995)
ABSTRACT
Multispectral, high-resolution imagery from the Advanced Very High Resolution Radiometer of NOAA polar
orbiting satellites is used to analyze the cloud-top structure of convective storms that develop a cirrus feature
above the anvil, referred to as a plume, whose origin remains unclear. Images from the radiometer’s channels
2, 3, and 4 and a combination of any two of these suggest a relationship between the emergence of such plumes
and a source of small ice particles (diameter around 3.7 mm, channel 3 wavelength) at the cloud top. Unique
observations of deep convective storms over Europe are presented and discussed. The paper does not provide
an exhaustive explanation of the phenomenon but contributes original material to the study of convective storm
cloud-top structure, which is far from being completely described.
1. Introduction
Significant cloud-top features, associated mostly
with severe storms, have been investigated by means
of geostationary meteorological satellites. Cold Ushaped and embedded warm spots, as well as cirrus
cloud layers above the anvil in the lower stratosphere,
have been observed by, among others, Fujita (1981,
1982), Negri (1982), Mack et al. (1983), and Heymsfield et al. (1991).
The geometric and spectral resolution of the Advanced Very High Resolution Radiometer (AVHRR)
on board National Oceanic and Atmospheric Administration (NOAA) polar orbiters lends itself well to the
detection of small-scale structures of storm tops not
adequately evinced by geostationary satellite imagery.
Liljas (1984) and Scorer (1986) first reported increased radiance in channel 3 (3.55–3.93 mm) on top
of some convective storms, recognizing the channel’s
potential for the identification of inhomogeneities in
their microphysical composition. Setvák and Doswell
(1991) report observations of deep convective storms
based on AVHRR’s channels 2 (0.725–1.1 mm), 3, and
4 (10.3–11.3 mm). Observations of convective storms
in NOAA/AVHRR imagery at the Czech Hydrometeorological Institute since 1984 have shown the exis-
tence of at least two major types of increased channel
3 cloud-top reflectivity (Setvák 1989): the ‘‘convective cell-like’’ and the much less frequent ‘‘plumelike.’’ The present paper focuses on the presence of
‘‘plumes’’ on top of some convective storms. These
structures clearly appear against the anvil background
in channel 2 or 3 and often in both. They originate from
a very small spot (the size of a few pixels) in the
storm’s embedded warm area and spread downwind,
resembling plumes from a chimney. Observations in
channel 2 show in most cases the vertical separation
between the plume and the underlying anvil. Observations of several storms in Europe and a discussion of
their possible origin are reported.
2. Observations
Two observations of plume-producing storms are detailed as a basis for discussion. The reader is also directed to Setvák and Doswell’s (1991) case of a mature
supercell storm on 18 August 1986 1330 UTC over the
Czech Republic. This storm shows all the typical satellite-observed features associated with supercell
storms: cold U-shape (horseshoe), close-in and distant
warm areas, and an extended anvil cloud.
a. Meteorological situation
Corresponding author address: Dr. Vincenzo Levizzani, Istituto
FISBAT–CNR, via Gobetti 101, I-40129 Bologna, Italy.
E-mail: [email protected]
Figure 1 shows a mature multicell storm on 6 July
1988 1348 UTC over southern Poland and north-central Slovakia as seen by NOAA-9. The surface chart at
361
q 1996 American Meteorological Society
v4483 / ams v4304
1279 Mp
361
Friday Oct 11 01:35 PM
AMS: J At Sci (February 1 96) 1279
362
JOURNAL OF THE ATMOSPHERIC SCIENCES
FIG. 1. 1348 UTC 6 July 1988: AVHRR NOAA-9 channel 2 image
of the multicellular storm over north-central Slovakia and southern
Poland (1400 km 1 1050 km). The box delimits the area covered by
all the subsequent images of the storm.
1200 UTC of the same day (not included) exhibits a
cold front that, encircling a moderately high pressure
center (1020 hPa) over eastern Germany and the Czech
Republic, extended from northeastern Italy across eastern Europe to the Baltic countries and then hooked
around to southern Sweden and Norway. On the border
between Poland and Slovakia, cumulus and cumulonimbus were reported with the formation of thunderstorms over a wide area; altocumulus were also observed from Russia down to Austria and Germany. The
system was in the mature stage at the time of satellite
observation, and at least two convective cores were active.
Figure 2 shows a multicell storm on 30 July 1994
0517 UTC over Normandy and the English Channel as
seen by NOAA-11 (biggest, elongated structure inland)
accompanied by the convective cluster of a companion
storm (offshore). The meteorological situation at 0000
UTC shows a depression over northwestern Europe,
which constitutes a typical summer event when the polar front along the westerlies extends down to relatively
low latitudes between 407 and 507N. The system was
driven by the wide low pressure area with its epicenter
of 985 hPa immediately south of Iceland. Along its
meridional border was a family of intensifying thunderstorms at various stages of development that was
moving toward central Europe. This was more or less
the case when the satellite passed over and documented
a series of active convective centers, besides the one
we are describing, over eastern France, Switzerland,
Germany, Belgium, the Netherlands, and southern
Denmark. Extensive precipitation was reported all over
central and eastern Europe resulting from the convergence of warm air from the SW and cold unstable air
from the NW. An analysis of Meteosat infrared (IR)
images detects a first storm around 2300 UTC on 29
July just south of Normandy and another storm forming
west of it at 2400 UTC. The storm cluster already in-
v4483 / ams v4304
1279 Mp
362
Friday Feb 07 02:28 PM
VOL. 53, NO. 3
cluded three storms by 0100 UTC on 30 July, and these
grew considerably bigger along a convergence line
stretching across Normandy to Brittany, where a new
storm formed around 0200 UTC. Around the time of
the NOAA passage, the storms were already massed
together, and the multicell was growing at a steady
pace, eventually acquiring the form of a supercell storm
(note that no radar observations are available so that
the inference of the multicell and supercell character is
entirely from satellite observation, chiefly IR). While
the storm west of it decayed around 0730 UTC, the
multicell storm seemed to reach maturity at 0800 UTC
and then proceeded in an ENE path to sweep first the
French coast of the English Channel and then the southeastern United Kingdom. The system disappeared by
1200 UTC.
b. Near and thermal infrared
Significant cloud-top structures usually associated
with supercell storms were first described by Fujita
(1981) on the basis of IR imagery from the Geostationary Operational Environmental Satellites (GOES).
A detailed overview of the observations of thunderstorm cloud-top surfaces can be found in, among others, Adler and Mack (1986), Heymsfield and Blackmer
(1988), and Heymsfield et al. (1991).
AVHRR channel 2 near-infrared observations allow
for a recognition of the spatial organization of the storm
at very high resolution: cell positioning, anvil extent,
wavelike patterns, and shadow casting. Although the
plumelike features are primarily observed in channel 3
imagery, it is possible in some cases to resolve them
simultaneously or detect them only in visible channel
1 (0.58–0.68 mm) or near-infrared channel 2. Thermal
IR cloud-top maps reveal the arrangement of convective cores, close-in and distant warm areas, and, under
certain circumstances, give an idea of the elevation map
FIG. 2. 0517 UTC 30 July 1994: AVHRR NOAA-11 channel 2
image of the multicellular storm over Normandy and the English
Channel (1400 km 1 1050 km). The box delimits the area covered
by all the subsequent images of the storm.
AMS: J At Sci (February 1 96) 1279
1 FEBRUARY 1996
L E V I Z Z A N I A N D S E T V Á K
of the top by means of combined radar and satellite
stereoscopic and IR observations as done by Negri
(1982) using GOES imagery.
A plume is detected in the channel 2 image of the 6
July 1988 storm (Fig. 3a), and its presence is evident
also from channel 4 (Fig. 3b), which in our experience
is very uncommon. The ‘‘source’’ of plumes is normally shifted downwind from the coldest areas, as it is
for this storm (Fig. 3c). Here, given the very narrow
width of the object at its very beginning in channel 2,
it is located in a pointlike area (a few kilometers in
diameter) within the warm spot of the big eastern cell.
The shadows being cast by the plume on the surrounding anvil top indicate its vertical separation from the
surrounding anvil, thus suggesting that the generating
mechanism is other than the anvil-producing one. One
more argument in favor of the separation, even though
not conclusive, is that the structure of the plume is more
or less preserved in channel 4, indicating a temperature
difference between the particles of the plume and the
surroundings. A vertical separation could well account
for this difference. Naturally, a different channel 4
emissivity of the smaller particles forming the plume
363
and the surrounding anvil is another candidate explanation.
The separation of the plume from the anvil top is
much more evident from the 30 July 1994 storm. The
height of the sun above the horizon is approximately
1.8 degrees, and the storm is situated at the center of
the satellite pass so that the sensor is looking at it nearly
from above. The very favorable sun–satellite–storm
position in the early morning produced larger shadows
in channel 2 (Fig. 4a), which show the plume body as
emerging from the storm’s main structure. Here, the
plume is more like a ‘‘split plume’’ with two very
bright splinters emerging from the source, which again
is to be found in the warm spot of an active cell (see
channel 4 image in Fig. 4b and diagram in Fig. 4c).
The two arms of the plume show up also in channel
4 as narrow discontinuities in the cloud-top thermal
field. The ‘‘stratospheric cirrus at the height of 15 km
in the form of two fingers . . . in the area of the distant
warm area’’ on GOES visible (VIS) imagery reported
by Mack et al. (1983) seems to be very similar to the
split plume of the 30 July 1994 storm and to that of the
storm in Setvák and Doswell’s (1991) Fig. 4. The cy-
FIG. 3. 1348 UTC 6 July 1988: (a) Channel 2 image; (b) channel 4 image with temperature range between 205 (red) and 255 K (blue);
(c) schematic diagram showing locations of cells, warm area of eastern cell, and plume (440 km 1 330 km).
v4483 / ams v4304
1279 Mp
363
Friday Feb 07 02:28 PM
AMS: J At Sci (February 1 96) 1279
364
JOURNAL OF THE ATMOSPHERIC SCIENCES
VOL. 53, NO. 3
FIG. 4. 0517 UTC 30 July 1994: (a) Channel 2 image; (b) channel 4 image with temperature range between 208 (red) and 228 K (violet),
a temperature of 204.5 K for the eastern overshooting top in the western storm (minimum temperature of the whole image), 208.5 K for the
overshooting top closest to the plume source, and 220.5 K the highest temperature within the warm spot; (c) schematic diagram showing
locations of eastern cell, its associated warm area, plume’s arms, and oval cloud (440 km 1 330 km).
clonic curvature of the northern branch and the anticyclonic one of the southern in the present case suggest
that some kind of opposing rotations are responsible,
together with the diverging cloud-top outflow, for the
plume’s forking. No inference whatsoever can be made
about the nature of these rotations as they originated
above the anvil or belonged to the ascending flow.
A bright spot in channel 2 in the form of an oval
cloud is detected in the same position as the plume’s
source. The shadow cast by this rounded cloud appears
much longer than that of the plume, suggesting that we
are looking at separate objects, perhaps at two different
heights. Very simple geometric computations based on
the shadow width give a height of about 300 m above
the anvil top for the oval cloud and a height of 50 m
for the plume’s north branch. No conclusions can be
drawn as to whether the source for both the oval cloud
and the plume is the same or not. An analysis of Meteosat imagery was carried out without success. Meteosat VIS images (0.5–0.9 mm) are available only from
0600 UTC: images at 0600 and 0630 UTC in single
and double resolution showed no trace of the plume,
nor did the IR Meteosat image at 0500 and 0530 UTC.
v4483 / ams v4304
1279 Mp
364
Friday Feb 07 02:28 PM
The reason for this is most surely the fact that Meteosat
spatial resolution is not high enough in the majority of
cases to detect small-scale features at the cloud top.
c. The 3.7-mm spectral band
AVHRR channel 3 is located in the spectral region
that includes both emitted and reflected components,
being at the long wavelength end of the solar spectrum
and at the short wavelength end of the blackbody emission spectrum at earth temperatures (Scorer 1986). The
channel 3 cloud-top appearance exhibits a complex dependence on several parameters for both the thermal
(temperature, emissivity, transparency) and reflected
(reflectivity, transparency, solar zenith angle, geometrical parameters affecting backscattering) components.
Many of these radiative properties are closely linked to
cloud-top microphysics and are strictly interconnected
(Hunt 1973; Scorer 1986). The channel is used for the
detection of ice particles at cloud top since they are
very absorbent at these wavelengths. Some convective
storms feature areas of increased outgoing channel 3
radiance (Liljas 1984, 1986), which is to be ascribed
AMS: J At Sci (February 1 96) 1279
1 FEBRUARY 1996
L E V I Z Z A N I A N D S E T V Á K
365
The plume of the 30 July 1994 storm is associated
with only a slight increase in channel 3 reflectivity (Fig.
6). The increased reflectivity coincides with the two
‘‘arms’’ delimiting the area of the plume and already
described for the channel 2 image. The oval cloud observed as a bright spot in channel 2 also exhibits increased channel 3 reflectivity, thus revealing a similar
composition with respect to the plume, a fact that could
not be proved by simple channel 2 and channel 4 analysis. Note, however, that due to the very low sun elevation angle, no exact quantitative estimate of channel
3 reflectivity can be performed in this case.
d. Multispectral analysis
FIG. 5. 1348 UTC 6 July 1988: Channel 3 cloud-top reflectivity
with values ranging from 0 to the 7.5% increase of the plume (440
km 1 330 km).
to the reflected component (Setvák 1989). Therefore,
the reflected component, specifically the channel 3
cloud-top reflectivity (Setvák and Doswell 1991), is to
be further investigated for a satisfactory interpretation
of the observed increased radiances.
The increased channel 3 cloud-top reflectivity is the
primary way to identify these objects, whose shape
clearly resembles a plume or a ‘‘fan’’ emanating from
a relatively small spot (the source), whose size is of
the order of a few pixel units (a few kilometers at
most). Another important characteristic is that their
length can stretch along the entire anvil and even beyond its margins without any significant intervening
decrease in reflectivity; Setvák and Doswell (1991)
have reported an example for a supercell storm (see
their Fig. 4). The mechanism responsible for the production of such a plume’s particles is necessarily a very
stable one in order to give rise to plumes of the observed length. Given upper-level relative winds of the
order of 10 to 25 m s 01 , the lifetime of the source is
of the order of tens of minutes, perhaps even several
hours.
The channel 3 cloud-top reflectivity image of the 6
July 1988 storm (Fig. 5) shows a strong increase in
reflectivity over the object identified as a plume in
channel 2 with a very sharp radiance gradient in the
southern flank. The most important difference between
the shape of the plume in channels 2 and 3 is at the
source, where the very beginning is missing in channel
3 but clearly detectable in channel 2. It seems that the
mechanism producing the plume, rather than ceasing
sometimes, has here simply undergone a modification
resulting in a different microphysical composition of
the plume, as suggested by the decrease in channel 3
reflectivity. Note that the plume is not split into two
beams, as is often the case, but has an intact body,
which probably reflects the environmental conditions
of the layer.
v4483 / ams v4304
1279 Mp
365
Friday Feb 07 02:28 PM
The complexity of the described cloud-top phenomena associated with deep convective storms requires
detailed multispectral analysis of AVHRR data in order
to extract information that would otherwise remain
concealed in single-channel analysis. The main goal is
to associate features observed in individual channels
and, consequently, to gain higher-order insights into a
storm’s cloud-top structure.
In most cases very accurate image processing reveals
shadows cast by the plumes on the anvil cloud top. This
proves that, on these occasions at least, the height of
the plume tops is greater than that of the anvil cloud
top. That shadows are not always detectable can be
attributed to either unfavorable geometrical arrangements of the sun–satellite–cloud-top system or the
characteristics of a particular plume (e.g., diffused
edges, high transparency, or the plume being embedded
in the anvil). When detected in channels 1 and 2, the
plumes are partially transparent. These facts, that is,
shadows and partial transparency, prove that plumes
constitute an independent upper-level structure of these
storms and therefore must be generated by mechanisms
other than the anvil-producing one. These observations
match Fujita’s (1982) on low-stratospheric cirrus
FIG. 6. 0517 UTC 30 July 1994: Original channel 3 image (no
quantitative channel 3 cloud-top reflectivity analysis was performed
due to very low sun elevation angle) (440 km 1 330 km).
AMS: J At Sci (February 1 96) 1279
366
JOURNAL OF THE ATMOSPHERIC SCIENCES
FIG. 7. 1348 UTC 6 July 1988: (a) Multispectral synthesis of channel 2 (yellow shades) and channel 3 cloud-top reflectivity (blue
shades); (b) enhanced channel 2 obtained by superimposing channel
4 color palette as in Fig. 3b. The plume appears evident in the mideastern section of the storm in channels 2 and 3 and a little less so
in channel 4. Its separation from the anvil top becomes more evident
in the multispectral images (440 km 1 330 km).
above the anvil and corroborate the cirrus’s separation
from the anvil’s structure as depicted in Fujita’s (1982)
Fig. 11. The plumes are normally undetectable in channels 4 and 5 and only on some occasions, like those of
the storms presented in this paper, do they increase the
anvil cloud-top temperature by 1 or 2 K.
Multispectral, color composite syntheses of at least
two channels into one image are commonly used in a
wide range of remote sensing applications. The color
of a particular object in such a composite image results
from the ratio of basic colors (red, green, and blue in
most cases) that are assigned to the individual spectral
channels combining in the synthesis. This technique
enables easier object recognition in the final image, as
the resultant colors depend on the multispectral characteristics of the objects. In meteorology, such techniques are mainly applied to cloud classification for
weather analysis and forecasting (e.g., Karlsson and
Liljas 1990).
v4483 / ams v4304
1279 Mp
366
Friday Feb 07 02:28 PM
VOL. 53, NO. 3
Channel 2 in Fig. 7a is associated with yellow
shades, and channel 3 cloud-top reflectivity with blue
ones. Relatively high channel 3 reflectivity of the
plume in conjunction with high albedo and texture in
channel 2 manifestly demonstrates that the plume is
overshooting the anvil top by very clearly delimiting
and giving depth to its southern flank. The channel 2
image of the same storm of 6 July 1988 is enhanced
by superposition of the channel 4 temperature field in
Fig. 7b: the result is the channel 2 structure colored
with the palette of channel 4. This makes it possible
unambiguously to observe that the plume’s temperature
is higher than the surrounding anvil’s, thereby situating
it higher up in the stratosphere. Moreover, the collocation of the plume’s source in channel 2 and the warm
spot in channel 4 are shown.
Channel 2 is associated with blue shades (the
brighter the shade, the higher the channel 2 albedo),
and channel 4 with red (the brighter the shade, the
colder the cloud top) for the 30 July 1994 storm in Fig.
8. The bifurcation of the plume in two arms is enhanced, thus confirming that it is an object separated
from the anvil body.
The very limited number of available cases documented by NOAA imagery stems from the relative
rareness of plume-generating storms over Europe, so
that their observations are very sparse. Twelve plumeproducing storms have been examined heretofore: five
had a cold U, and the plume originated from either a
close-in or distant warm area; four did not show any U
(or V), and the plume still emanated from a warm spot;
and in the last three the plumes were not connected to
any significant structure in channel 4, and their source
was in the region downwind from the storm’s core. The
synthesis of channels 3 (blue) and 4 (red) of the 18
August 1986 storm [see Setvák and Doswell (1991)
for single-channel imagery] in Fig. 9 clearly identifies
the distant warm area as the source of the split plume.
FIG. 8. 0517 UTC 30 July 1994: Multispectral synthesis of
channels 2 (blue shades) and 4 (red shades) (440 km 1 330
km).
AMS: J At Sci (February 1 96) 1279
1 FEBRUARY 1996
L E V I Z Z A N I A N D S E T V Á K
FIG. 9. 1330 UTC 18 August 1986: Supercell storm over northcentral Bohemia. Synthesis of channel 3 cloud-top reflectivity (blue)
and channel 4 (red). The multispectral image shows that the plume
emanates from the channel 4 distant warm area [the reader is directed
to Setvák and Doswell (1991) for single-channel imagery] (440 km
1 330 km).
From an analysis of convective storms in which the
well-pronounced warm–cold couplet appears simultaneously with channel 3 plumes, it follows that the
plumes emanate from warm spots located within storm
cores.
Synthesis of channel 2 and channel 4 data in some
cases indicates that a dark spot in channel 2 not associated with any nearby overshooting top might be collocated with the channel 4 warm spot (emanating a
plume visible in channel 3). This can either correspond
to the presence of a cloud-top slope or turret (McKee
and Klehr 1978) or a significantly lower ice particle
concentration. However, the relative positioning of the
sun–satellite–cloud-top system and the highly variable
cloud-top topography, which markedly affect channels
1 and 2 data, usually make these images difficult to
interpret unambiguously.
Plumelike features are occasionally detected in channel 1 or channel 2 imagery, with no correspondence in
channel 3. It remains to be seen whether such events
are at all related to the plumes in channel 3, differing
simply as to particle size, shape, or orientation, or represent a completely different phenomenon, perhaps a
wavelike disturbance.
3. Discussion
Very small ice particles, whose size is comparable
or somewhat larger than channel 3 wavelength, are very
likely to increase channel 3 cloud-top reflectivity significantly (Hunt 1973; Arking and Childs 1985). The
observation of enhanced channel 3 reflectivity throughout a storm’s anvil indicates that the ice particles responsible for the increase in reflectivity at the cloud top
persist there without major changes for a long time.
While ice crystal sizes in cirrus clouds of nonconvective origin range between several tens and hundreds
v4483 / ams v4304
1279 Mp
367
Friday Feb 07 02:28 PM
367
of microns (Pruppacher and Klett 1978), the observed
increased channel 3 cloud-top reflectivity values in
convective storm tops suggest the presence of ice particles whose size is to be gauged in units of microns
(size of the order of channel 3 wavelength). There
might be two alternative processes leading to these increased small ice particle concentrations. First, the
growth time of ice particles might be strongly limited
as a result of violent vertical transport by analogy with
the bounded weak echo region in radar observations of
strong updrafts (Burgess and Lemon 1990). Second,
the vertical lifting and/or gravitational settling might
well separate the smallest particles at the cloud top.
These mechanisms are currently beyond our verification capabilities and require observations by other techniques (e.g., airborne radar, in situ microphysical measurements).
Within the cloud-top temperature range of deep convective storms, only ice particles can be expected at the
storm’s top (Detwiler et al. 1992), although no conclusive information from in situ measurements is available at present on very small particle distribution ( õ10mm size) inside severe storm cloud tops (Heymsfield
1986; Detwiler et al. 1992). Straka (1989) reports a
considerable concentration of small ice particles at the
anvil level in the simulation of a slowly evolving multicell storm over northeastern Colorado by means of
the Wisconsin dynamical/microphysical model. The
particles that formed in and near the updraft core grew
rapidly in size before being transported upward and
exhausted into the anvil, although a large number of
them still remained at this level. The simulation of the
2 August 1981 Cooperative Convective Precipitation
Experiment (CCOPE) supercell storm by means of the
same model (Johnson et al. 1993, 1994) produces ice
particles of 4–5-mm diameter in the bounded weak
echo region (P. K. Wang 1994, personal communication), in good agreement with the observation of
6-mm particles in the same volume.
Having noted the presence of cirrus 1–3 km above
the anvil top in the lower stratosphere, whose temperature was then higher than that of the anvil cirrus, Fujita
(1982) proposed the following mechanism. The radiance emitted by the anvil cloud top can be increased
by the thin, warmer stratospheric cirrus layer, wherein
particle concentrations are higher in the wake (referring to the storm-relative upper-level winds) of the
overshooting top. The generation of the stratospheric
cirrus [photograph in Fig. 11 of Fujita (1982)] was
attributed to the perturbation caused by violent collapses of the overshooting tops into the wake region.
No satisfactory mechanism for plume production can
be put forward on the basis of these observations alone.
The following considerations based on current findings,
however, may lead to further investigations for a better
understanding of the phenomenon.
• Plume formation and development as part of anvil
growth seems unlikely at the present state of knowl-
AMS: J At Sci (February 1 96) 1279
368
JOURNAL OF THE ATMOSPHERIC SCIENCES
edge. The present paper’s channel 2 observation of
shadows cast by plumes on the anvil background indicates their mutual separation. Heymsfield’s (1986)
results on ice particle evolution in the anvil of a severe
thunderstorm during CCOPE show that aggregation is
an important mechanism for particle growth in the anvil, where conditions are favorable for aggregation
even at temperatures much below 0207C. With aggregation, the maximum particle size increases along with
distance downwind of the updraft core. In the case of
plumes, however, the increased channel 3 reflectivity
remains more or less constant at great distance from the
plume’s source. It is reasonable to assume that the
mechanism of plume formation and development is
other than the ones involved in the anvil cirrus production.
• The plumes described in the previous sections
show characteristics that identify them as different
from Fujita’s (1982) lower-stratospheric cirrus. First,
they cover a well-defined area of the anvil downwind
region and never occupy the entire anvil area. Second,
they do not jump back and cover the overshooting top,
as Fujita’s (1982) do. The last, and probably most important, characteristic is their very small source (a few
pixels in diameter). Located far behind the overshooting top, which can hardly represent the location of repeated violent collapses of the air from the dome, it is
too constant both in space and time.
• The mechanism of subsidence and adiabatic
warming for the formation of the close-in warm area
proposed by Heymsfield et al. (1983) provides no clues
as to the plume’s formation. Their hypothesis, however, that the airflow over the overshooting top would
resemble that over a mountain with consequent subsidence and damped oscillations presents another inference. The formation of plumes as a wavelike phenomenon in the lee of the overshooting top seems unlikely
due to both the temporal stability of the plumes and the
absence of any oscillations that would be clearly visible
given the AVHRR resolution, which is much greater
than the GOES IRs of Heymsfield et al. (1983). In
other cases we have observed wave clouds on top of
convective storms in the wake of overshooting tops or
gravity waves related to pulsating updrafts, but in none
of these was an increased channel 3 reflectivity observed.
Plume development was observed in one case (27
September 1992, not shown) for 2 h on Meteosat VIS
imagery. The plume and the warm spot originated,
evolved, and disappeared almost simultaneously,
thereby suggesting the presence of a dynamic link between the two phenomena. This fact, together with the
close link between plume source position and warm
spot location, suggests that the theory of warm spot
formation must include a mechanism for plume production under certain circumstances. The rapid scan
mode of recent GOES-I carrying a 3.9-mm channel
v4483 / ams v4304
1279 Mp
368
Friday Feb 07 02:28 PM
VOL. 53, NO. 3
(Menzel and Purdom 1994) will, it is hoped, provide
more detailed and timely observations of this link.
Acknowledgments. David W. Martin, Eric A. Smith,
and anonymous reviewers deserve special thanks for
their invaluable suggestions, which greatly improved
the paper. The study was funded by the Czech Ministry
of the Environment, Agenzia Spaziale Italiana, Gruppo
Nazionale per la Difesa dalle Catastrofi Idrogeologiche,
and Piano Finalizzato Sistemi Informatici e Calcolo
Parallelo. One of the authors (MS) was supported by
the European Communities as part of the Community’s
Action for Cooperation in Science and Technology
with Central and Eastern European Countries, under
Contract ERB 3510PL920837.
REFERENCES
Adler, R. F., and R. A. Mack, 1986: Thunderstorm cloud top dynamics as inferred from satellite observations and a cloud top parcel
model. J. Atmos. Sci., 43, 1945–1960.
Arking, A., and J. D. Childs, 1985: Retrieval of cloud cover parameters from multispectral satellite images. J. Climate Appl. Meteor., 24, 322–333.
Burgess, D. W., and L. R. Lemon, 1990: Severe thunderstorm detection by radar. Radar in Meteorology, D. Atlas, Ed., Amer. Meteor. Soc., 619–656.
Detwiler, A. G., P. L. Smith, and J. L. Stith, 1992: Observations of
microphysical evolution in a High Plains thunderstorm. Proc.
11th Int. Conf. on Clouds and Precipitation, Vol. 1, Montreal,
PQ, Canada, International Commission on Clouds and Precipitation and International Association of Meteorology and Atmospheric Physics, 260–263.
Fujita, T. T., 1981: Mesoscale aspects of convective storms. Proc.
IAMAP Symp. on Nowcasting: Mesoscale Observations and
Short-Range Prediction, Hamburg, Germany, European Space
Agency, 3–10.
, 1982: Principle of stereoscopic height computations and their
applications to stratospheric cirrus over severe thunderstorms.
J. Meteor. Soc. Japan, 60, 355–368.
Heymsfield, A. J., 1986: Ice particle evolution in the anvil of a severe
thunderstorm during CCOPE. J. Atmos. Sci., 43, 2463–2478.
Heymsfield, G. M., and R. H. Blackmer Jr., 1988: Satellite-observed
characteristics of Midwest severe thunderstorm anvils. Mon.
Wea. Rev., 116, 2200–2224.
, R. H. Blackmer Jr., and S. Schotz, 1983: Upper-level structure
of Oklahoma tornadic storms on 2 May 1979. I: Radar and satellite observations. J. Atmos. Sci., 40, 1741–1755.
, R. Fulton, and J. D. Spinhirne, 1991: Aircraft overflight measurements of midwest severe storms: implications on geosynchronous satellite interpretations. Mon. Wea. Rev., 119, 436–
456.
Hunt, G. E., 1973: Radiative properties of terrestrial clouds at visible
and infra-red thermal window wavelengths. Quart. J. Roy. Meteor. Soc., 99, 346–369.
Johnson, D. E., P. K. Wang, and J. M. Straka, 1993: Numerical simulations of the 2 August 1981 CCOPE supercell storm with and
without ice microphysics. J. Appl. Meteor., 32, 745–759.
,
, and
, 1994: A study of microphysical processes in
the 2 August 1981 CCOPE supercell storm. Atmos. Res., 33,
93–123.
Karlsson, K.-G., and E. Liljas, 1990: The SMHI model for cloud and
precipitation analysis from multispectral AVHRR data.
PROMIS Rep. 10, Swedish Meteorological and Hydrological
Institute, Norrköping, Sweden, 74 pp.
Liljas, E., 1984: Processed satellite imageries for operational forecasting. Swedish Meteorological and Hydrological Institute,
Norrköping, Sweden, 43 pp.
AMS: J At Sci (February 1 96) 1279
1 FEBRUARY 1996
L E V I Z Z A N I A N D S E T V Á K
, 1986: Use of the AVHRR 3.7 micrometer channel in multispectral cloud classification. PROMIS Rep. 2, Swedish Meteorological and Hydrological Institute, Norrköping, Sweden, 23
pp.
Mack, R. A., A. F. Hasler, and R. F. Adler, 1983: Thunderstorm cloud
top observations using satellite stereoscopy. Mon. Wea. Rev.,
111, 1949–1963.
McKee, T. B., and J. T. Klehr, 1978: Effects of cloud shape on scattered solar radiation. Mon. Wea. Rev., 106, 399–404.
Menzel, W. P., and J. F. W. Purdom, 1994: Introducing GOES-I: The
first of a new generation of geostationary operational environmental satellites. Bull. Amer. Meteor. Soc., 75, 757–781.
Negri, A. J., 1982: Cloud-top structure of tornadic storms on 10 April
1979 from rapid scan and stereo satellite observations. Bull.
Amer. Meteor. Soc., 63, 1151–1159.
v4483 / ams v4304
1279 Mp
369
Friday Feb 07 02:28 PM
369
Pruppacher, H. R., and J. D. Klett, 1978: Microphysics of Clouds and
Precipitation. D. Reidel, 714 pp.
Scorer, R. S., 1986: Cloud Investigation by Satellite. Ellis Horwood
Ltd., 314 pp.
Setvák, M., 1989: Convective storms—The AVHRR channel 3 cloud
top reflectivity as a consequence of internal processes. Proc.
Fifth WMO Conf. on Weather Modification and Applied Cloud
Physics, Beijing, Peoples’ Republic of China, World Meteor.
Org., 109–112.
, and C. A. Doswell III, 1991: The AVHRR channel 3 cloud top
reflectivity of convective storms. Mon. Wea. Rev., 119, 841–
847.
Straka, J. M., 1989: Hail growth in a highly glaciated central high
plains multi-cellular hailstorm. Ph.D. dissertation, University of
Wisconsin—Madison, 413 pp.
AMS: J At Sci (February 1 96) 1279